XA_EMB
Embodied Multimodal Brain.
An artificial intelligence model designed to be the brain of a robot. It learns and works in a way modelled on the human brain.
This section is still in development.
XA_EMB is in an early stage of development. There is nothing to download or test yet. This page will be updated as the work progresses.
What it is
A brain for robots, not a chatbot.
Most AI models are trained once and then stop learning. XA_EMB is designed to keep learning from what the robot sees, hears and does, the same way a person learns from the world around them.
Overview
- Full name
- Embodied Multimodal Brain
- Type
- AI model
- Purpose
- The brain of a robot
- Works with
- XA_Robotics
- Status
- In development
The name
Three words, one idea.
Embodied. The model lives inside a physical body, a robot, and learns through that body.
Multimodal. It takes in several kinds of input together, such as images, sound and movement, instead of text alone.
Brain. It works the way a human brain does: it learns from experience, keeps what it has learned and uses it to decide what to do next.
Design goals
What XA_EMB is being built to do.
These goals guide development. They describe the direction of the work, not finished features.
Perceive like a person. Combine sight, sound, touch and the position of the body into one understanding of the world, instead of handling each sense separately.
Learn by doing. Learn from acting in the physical world and from the results of those actions, not only from data collected in advance.
Keep learning. Continue learning after it is deployed, adding new skills without forgetting the ones it already has.
Built to help. Give the robots of XA_Robotics a brain that can work safely next to people and assist them in everyday tasks.